SUPPLEMENTARY FIGURE LEGENDS
Figure S1. Graphical representation of hypothetical evolutionary scenarios of medfly using DIYABC-RF. The strategy focused on testing hypotheses of colonisation routes from South Africa to Brazil or South Africa to the rest of the sampling sites. A) Analysis 1, in light grey, highlights the best fit Scenario. B) Analysis 2, using only Scenarios 2, 4 and 5 (the highest CV values in Analysis 1). The best fit scenario is highlighted in light blue. CV: corresponded to the classification vote for each scenario obtained in the different analyses.
Figure S2. Pairwise Fst values for the six populations of medfly. All calculations were significant, abbreviations are based on where populations are collected. SA, South Africa; SP, Spain; GR, Greece; GU, Guatemala; BR, Brazil; AU, Australia.
Figure S3. AIC statistic plots where lower value indicates optimal clustering. A) Optimal AIC estimation using the dataset with the six populations. B) Optimal AIC estimation using only populations collected in introduced range (i.e. SP, GR, GU, BR, AU). Delta K estimation by Evanno method. C) Best delta K estimation using the dataset with the six populations. D) Best delta K estimation using only populations collected in introduced range (i.e. SP, GR, GU, BR, AU).
Figure S4. Clustering analysis using the 1907 SNPs. A) Group assignment probability plot using K=3 based on DAPC analysis. The colour of cells points out the probability of assigning a given sample (red: high probability) and the blue crosses indicate the true group assignment. The row labels are specimens ID and columns (clusters) correspond to the population groups. B) Hierarchical clustering (Ward clustering), SA: South Africa, BR: Brazil, OP: Other populations. Dark red arrow is highlighting the specimen SA_8 assigned to the major cluster formed by populations collected in the introduced region.
Figure S5. Projection of dataset (observed data) from the training set on Linear Discriminant Analysis plots. C) Analysis 1, six scenarios analysed individually. D) Analysis 2, three scenarios analysed individually.
Figure S6. Tree cloud produced by DENSITREE from SNAPP analysis. Left: tree cloud of the 15 consensus trees of Fig. 6.Right : tree cloud obtained by independent subsample set from the six populations studied.  Maximum-clade-credibility tree estimation is shown in dark blue (most highly supported), red is the next most supported, and green is the least supported. Maximum-clade-credibility tree shown in the black right-angled tree with posterior probabilities at nodes. Branch width is proportional to theta. 
Figure S7. Plot of sequencing read depth for microbiota database and the rarefaction curve.
Figure S8. Number of unique and shared ASVs across different biogeographical regions. A) South Africa compared to Palearctic sampling locations. B) South Africa compared to Neotropical sampling locations. C) South Africa compared to Australia (Australasian).
Figure S9. Alpha diversity of medfly microbiome across six sampling locations. Observed (number of ASVs present -species richness), Shannon, Inverse Simpson and Pielou’s evenness diversity indexes.